ollama local models cannot support

[RunningDW]

tried qwen3-vl:8b or gemma3:12b, always cannot get connected

[arkml]

Looking into this.

[arkml]

qwen3-vl:8b seems like a vision lm and gemma3:12b has a different approach to tool calling which is likely why you see the issue. I'm able to replicate the issue with them.

A model similar to gpt-oss:20b should work well.